Reliability of DSM-IV Anxiety and Mood Disorders: Implications for the Classification of Emotional Disorders

Transcription

1 Journal of Abnormal Psyhology 200, Vol. 0, No.,49-58 Copyright 200 by the Amerian Psyhologial Assoiation, In X/0/S5.00 DOI: 0.037// X Reliability of DSM-IV Anxiety and Mood Disorders: Impliations for the Classifiation of Emotional Disorders Timothy A. Brown Boston University Peter A. Di Nardo State University of New York at Oneonta Cassandra L. Lehman and Laura A. Campbell Boston University The reliability of urrent and lifetime Diagnosti and Statistial Manual of Mental Disorders (4th ed.; DSM-IV; Amerian Psyhiatri Assoiation, 994) anxiety and mood disorders was examined in 362 outpatients who underwent 2 independent administrations of the Anxiety Disorders Interview Shedule for DSM-IV: Lifetime version (ADIS-IV-L). Good to exellent reliability was obtained for the majority of DSM IV ategories. For many disorders, a ommon soure of unreliability was disagreements on whether onstituent symptoms were suffiient in number, severity, or duration to meet DSM-IV diagnosti riteria. These analyses also highlighted potential boundary problems for some disorders (e.g., generalized anxiety disorder and major depressive disorder). Analyses of ADIS-IV-L linial ratings (0-8 sales) indiated favorable interrater agreement for the dimensional features of DSM-IV anxiety and mood disorders. The findings are disussed in regard to their impliations for the lassifiation of emotional disorders. Classifiation of emotional disorders has been an inexat siene, refleted by the modest reliability of many diagnosti ategories and marked hanges in definitional riteria aross editions of the Diagnosti and Statistial Manual of Mental Disorders (DSM; Amerian Psyhiatri Assoiation, 987, 994). The diagnosti riteria for all anxiety and mood disorders were revised to varying degrees in the urrent, fourth edition of the DSM (DSM- IV; Amerian Psyhiatri Assoiation, 994). Often, these revisions were guided by reliability findings from large-sale studies of disorders from the revised, third edition of the DSM (DSM- III-R; Amerian Psyhiatri Assoiation, 987; see Di Nardo, Moras, Barlow, Rapee, & Brown, 993; Mannuzza et al., 989; Williams et al., 992). For example, in addition to the introdution of a formal typology of pani attaks (i.e., unexpeted, situationally predisposed, situationally bound; f. Barlow, Brown, & Craske, 994), DSM-IV riteria for pani disorder and agoraphobia no longer inlude severity speifiers (i.e., mild, moderate, severe). This revision was based on findings that whereas generally good interrater onsisteny was noted for dimensional india- Timothy A. Brown, Cassandra L. Lehman, and Laura A. Campbell, Center for Anxiety and Related Disorders, Boston University; Peter A. Di Nardo, Department of Psyhology, State University of New York at Oneonta. We thank Bonnie Conklin, Patriia Miller, Jeanne Esler, Jonathan Lerner, Jessia Grisham, and David Barlow for their assistane with this study. Correspondene onerning this artile should be addressed to Timothy A. Brown, Center for Anxiety and Related Disorders, Boston University, 648 Beaon Street, 6th floor, Boston, Massahusetts Eletroni mail may be sent to tabrown@bu.edu. 49 tors of pani frequeny and agoraphobia severity (e.g., in Di Nardo et al., 993, the orrelation between independent dimensional ratings of agoraphobi avoidane was.8), appliation of these ategorial severity speifiers was assoiated with onsiderable unreliability. For instane, in Di Nardo et al. (993), higher reliability was observed for urrent DSM-III-R pani disorder ollapsing aross all levels of agoraphobi avoidane («=.7) than for eah level of agoraphobia severity (xs =.6,.70,.40, for mild, moderate, and severe agoraphobia, respetively). Indeed, it has been found that diagnosti unreliability of DSM disorders often does not stem from disagreement on the presene of defining symptoms but rather from diffiulties applying ategorial utoffs to these inherently dimensional phenomena (e.g., DSM threshold for presene or absene of disorder based on suffiient distress or lifestyle impairment; appliation of DSM severity or ourse speifiers). In Di Nardo et al. (993) many of the diagnosti disagreements involving soial phobia and speifi phobia were ases in whih both interviewers noted lear features of these disorders but did not onur that these symptoms met the DSM-III-R interferene or distress threshold (f. Antony et al., 994; Stein, Walker, & Forde, 994). Another important issue in the lassifiation of emotional disorders is the diagnosti reliability of generalized anxiety disorder (GAD). In large sale studies entailing administration of two independent strutured interviews, DSM-IH-R GAD was assoiated with poor to fair reliability (kappas for urrent GAD were.27 in Mannuzza et al., 989,.53 in Di Nardo et al., 993, and.56 in Williams et al., 992). These findings, along with data indiating that GAD has a omorbidity rate exeeding 80% (e.g., Brawman- Mintzer et al., 993; Brown & Barlow, 992), led to debate among researhers as to whether there was suffiient evidene of disrimi-

2 50 BROWN, Dl NARDO, LEHMAN, AND CAMPBELL nant validity to retain GAD as a diagnosti ategory in DSM-IV (Brown, Barlow, & Liebowitz, 994). Although GAD remains a formal ategory in DSM-IV, its diagnosti riteria were revised substantially in an effort to define its boundary in relation to mood and adjustment disorders, anxiety disorders, and nonpathologial worry. These revisions inlude the requirement that worry must be pereived by the person as unontrollable (based on evidene that the parameter of unontrollability distinguishes GAD worry from normal worry; Abel & Borkove, 995; Borkove, 994) and the redution in the number of symptoms forming the assoiated symptom riterion from 8 to 6 (symptoms of autonomi arousal were eliminated [e.g., aelerated heart rate, shortness of breath]; symptoms of tension and negative affet were retained [e.g., musle tension, feeling keyed up/on edge, irritability]). Although the deision to eliminate autonomi symptoms was data driven (Brown, Marten, & Barlow, 995; Marten et al., 993), researhers have raised onern that this revision may obfusate the boundary between GAD and the mood disorders (Clark & Watson, 99). This boundary issue is refleted in a DSM-IV exlusionary riterion stating that GAD should not be assigned if its features our exlusively during the ourse of a mood disorder. Nonetheless, it is important to determine whether the substantial hanges to GAD in DSM-IV have resulted in improved diagnosti reliability. Similarly, it would be of interest to evaluate what impat other modifiations to the diagnosti definitions of emotional disorders have had on their reliability. Although the ategory of speifi phobia has historially been assoiated with favorable interrater agreement (e.g., KS >.80 in Di Nardo et al., 993, and Mannuzza et al., 989), D5M-/Vnow requires that this diagnosis be assigned as one of the following types: (a) animal (e.g., dogs, rats); (b) natural environment (e.g., heights, storms); () blood/injury/injetion (e.g., having a blood test); (d) situational (e.g., driving, enlosed plaes); and other (e.g., illness, vomiting). Although speifiation of speifi phobia types was intended to aount for the heterogeneity of the disorder, researh is needed on the reliability and validity of these distintions (f. Antony, Brown, & Barlow, 997). The purpose of this study was to evaluate the reliability and fators ontributing to diagnosti disagreements of the DSM-IV anxiety and mood disorders using the Anxiety Disorders Interview Shedule for DSM-IV: Lifetime version (ADIS-IV-L; Di Nardo, Brown, & Barlow, 994). The revisions in the ADIS-IV-L go well beyond updating the Anxiety Disorders Interview Shedule Revised (ADIS-R; Di Nardo & Barlow, 988) to be onsistent with DSM-IV riteria. Unlike the ADIS-R, the ADIS-IV-L provides diagnosti assessment of a broader range of onditions (e.g., substane use disorders), evaluation of lifetime disorders, and dimensional assessment of the key and assoiated features of disorders, irrespetive of whether a formal DSM-IV diagnosis is under onsideration (see Method setion). The latter revision is based on the position that many features of emotional disorders operate on a ontinuum rather than in a ategorial, presene/ absene fashion as in DSM diagnosis (f. Brown, 996; Brown, Chorpita, & Barlow, 998; Costello, 992). Beause of the importane of these dimensional ratings as indiators in linial trials and nosology and psyhopathology studies (e.g., Borkove & Costello, 993; Brown et al., 998), another aim of this study was to examine the interrater reliability of these measures. Partiipants Method Partiipants were 362 patients presenting for assessment and treatment at the Center for Stress and Anxiety Disorders, University at Albany, State University of New York («= 70), and the Center for Anxiety and Related Disorders, Boston University (n = 292).' (The two researh enters are olletively referred to as "the enter.") Women onstituted the larger portion of the sample (58%); average age was 33. (SD = 0.62, range = 8 to 62). The raial and ethni breakdown of the sample was Cauasian (88%), Afrian Amerian (4%), Hispani (3%), Asian (3%), Paifi Islander (%), and other or missing (2%). Patients were required to meet several inlusion and exlusion riteria that were assessed by telephone sreening at initial ontat with the enter and reassessed and onfirmed during the diagnosti interviews. Speifially, patients were required to be between the ages of 8 and 65 and to have a presenting omplaint that likely involved an anxiety or mood disorder. Patients were exluded from the study if any of the following were present: (a) urrent halluinations or delusions, (b) urrent or reent (within the past 6 months) alohol or substane abuse or dependene, () urrent suiidal or homiidal risk meriting risis intervention, and (d) two or more hospitalizations in the past 5 years for psyhoti symptoms. Patients were also required to meet psyhotropi mediation and psyhotherapy stabilization riteria for the periods preeding and overlapping with the diagnosti assessment. Patients using anxiolytis and beta-blokers were required to maintain the same dosage for at least month. Patients on antidepressants (triylis, seletive serotonin reuptake inhibitors, and monoamine oxidase inhibitors) had to maintain a stable dosage for at least 3 months. The mediation wash out period (i.e., period sine mediation disontinuation) was month for all mediations. Patients in psyhotherapy for an emotional problem were required to satisfy a 3-month stabilization period; the psyhotherapy wash out period was month. The urrent sample was randomly seleted to reeive two independent ADIS-IV-L interviews from roughly,400 onseutive admissions to the enter who met eligibility riteria between the periods of Deember 994 and Otober 999. In most ases (79%), the seond ADIS-IV-L ourred within 2 weeks of the first interview (M = 0.60 days, SD = 8.60). After both interviews had been ompleted and the interviewers had independently reorded their diagnosti judgments, ases were presented in weekly staff meetings that entailed the presentation of interviewers' diagnoses, disussion of fators ontributing to any diagnosti disagreements, and establishment of onsensus diagnoses. The primary soure of unreliability for eah diagnosti disagreement was reorded (by Timothy A. Brown or Peter A. Di Nardo) using a rating system designed for use in the present study: (a) differene in report patient gives different information to the two interviewers (e.g., variability in responses to inquiry about the presene, severity, or duration of key symptoms); (b) threshold onsistent symptom report is provided aross interviews, but interviewers disagree on whether these symptoms ause suffiient interferene and distress to satisfy the DSM-IV threshold for a linial disorder; () hange in linial status lear hange in the severity or presene of symptoms between interviews; (d) interviewer error interviewer improperly applies DSM-IV diagnosti or exlusion rules or fails to obtain neessary diagnosti information during ADIS-rV-L administration (e.g., skips an ADIS-IV-L diagnosti setion prematurely); (e) diagnosis subsumed under another ondition disagreement on whether symptoms are attributable to, or better aounted for by, a o-ourring disorder; and (f) DSM-IV inlarity disagreement stems from limitations of the DSM-IV riteria in providing lear diretion for differential diagnosis. Our researh enter reloated from the University at Albany, State University of New York, to Boston University in September 996.

3 RELIABILITY OF DSM-IV ANXIETY AND MOOD DISORDERS 5 Anxiety Disorder Interview Shedule for DSM-IV: Lifetime Version (ADIS-IV-L; Di Nardo et al, 994) The ADIS-IV-L is a semistrutured interview designed to establish reliable diagnosis of the DSM-IV anxiety, mood, somatoform, and substane use disorders and to sreen for the presene of other onditions (e.g., psyhoti disorders). The ADIS-IV-L is a substantial revision of the ADIS-R. In addition to being updated for DSM-IV riteria, the ADIS- IV-L provides assessment of lifetime disorders and a diagnosti timeline that fosters aurate determination of the onset, remission, and temporal sequene of urrent and lifetime disorders. Moreover, in several ADIS- IV-L setions, raters make dimensional ratings (0-8) of disorder features regardless of whether a DSM-IV diagnosis is under onsideration. This ours in the following setions: (a) soial phobia ratings of fear or avoidane of 3 soial situations; (b) generalized anxiety disorder ratings of exessiveness and diffiulty ontrolling worry in 8 areas; () obsessiveompulsive disorder ratings of persistene, distress, and resistane of 9 obsession types and frequeny of 6 ompulsions; and (d) speifi phobia ratings of fear or avoidane of 7 objets or situations from the 5 types of DSM-IV speifi phobias (animals, natural environment, blood/injetion/ injury, situational, other). Dimensional ratings of the features of pani disorder and agoraphobia are ompleted by interviewers only if these diagnoses are under onsideration (otherwise, the interviewer would skip this diagnosti setion after reeiving negative responses to initial sreening questions). Ratings in the pani disorder and agoraphobia setions inlude (a) frequeny of pani attaks in the past month, (b) fear of pani attaks in the past month (0-8 sale), and () urrent avoidane of or esape from 22 agoraphobi situations (0-8 sale). Dimensional ratings (0-8 sales) in the major depression and dysthymia setions and the assoiated symptoms portion of the generalized anxiety disorder setion are arranged in the same fashion as the pani disorder and agoraphobia setions of the ADIS-IV-L. However, for purposes of the present and other ongoing studies, in the Boston University sample (n = 292), interviewers inquired about and assigned these ratings regardless of whether a mood or generalized anxiety disorder diagnosis was under onsideration. These ratings were as follows: (a) major depression ratings of the seven symptoms that aompany depressed mood and diminished interest and pleasure in ativities to form the key riterion of major depressive episode; (b) dysthymia ratings of the six symptoms omprising its assoiated symptom riterion; and () generalized anxiety disorder ratings of the frequeny and severity of the six symptoms omprising its assoiated symptoms riterion. In these and other ADIS IV-L setions, interviewers followed the appropriate DSM-FV duration riterion (e.g., more days than not for a period of 2 years or greater in dysthymia) when making dimensional ratings (i.e., ratings refleted a omposite of severity, frequeny, or duration in respet to the DSM-IV riterion, if speified). For eah urrent and lifetime diagnosis, interviewers assigned a 0-8 linial severity rating that indiated their judgment of the degree of distress and interferene in funtioning assoiated with the disorder (0 = none to 8 = very severely disturbing/disabling). In instanes in whih the patient met riteria for two or more urrent diagnoses, the prinipal diagnosis was the one that reeived the highest linial severity rating. For both urrent and lifetime disorders, those that met DSM-IV riteria for a formal diagnosis were assigned linial severity ratings of 4 (definitely disturbing/disabling) or higher (linial diagnoses). Current linial diagnoses that were not deemed to be the prinipal diagnosis are referred to as additional diagnoses. When the key features of a urrent or lifetime disorder were present but were not judged to be extensive or severe enough to warrant a formal DSM IV diagnosis (or for DSM IV disorders in partial remission), linial severity ratings of -3 were assigned (sublinial diagnoses). When no features of a disorder were present, linial severity ratings of 0 were given. Interviewers Diagnostiians were 6 dotoral-level linial psyhologists and 30 advaned linial dotoral students. Before partiipating in the study, diagnostiians were required to undergo extensive training and meet strit ertifiation riteria in the administration of the ADIS-IV-L. Training began with the trainees reading the ADIS-IV L manual, observing videotaped interviews, and then observing at least three live ADIS-IV-L interviews onduted by a senior, ertified interviewer. While observing live interviews, the trainee made ratings and diagnoses. After the interview, the trainee and senior interviewer ompared and disussed diagnoses and dimensional ratings. Following observation of several live interviews, trainees had the option to administer one or more ollaborative interviews to beome more omfortable with ADIS-IV-L administration prior to the ertifiation phase. In a ollaborative interview, the trainee assumed primary responsibility for ADIS-IV-L administration, but the senior interviewer ould interjet as needed (e.g., ask differential diagnosis questions the trainee had not asked or provide an indiation of when to skip a diagnosti setion). In the ertifiation phase, trainees were required to administer a minimum of three ADIS-IV-Ls under observation of a senior interviewer. After the interview, the trainee and senior interviewer independently established urrent and lifetime diagnoses. The riteria for ADIS-IV-L ertifiation was that within three of five onseutive interviews, the trainee's diagnoses must math the senior interviewers' diagnoses and the trainee must ommit no ADIS-IV-L administration errors based on a heklist of nine items (e.g., omission of mandatory inquiry or failure to ask neessary follow-up questions of larifiation). A math was defined as (a) agreement on the prinipal diagnosis (inluding DSM-IV severity desriptors suh as major depression, single episode, moderate) and agreement within point on its linial severity rating and (b) identifiation as a linial disorder all additional and lifetime diagnoses assigned by the senior interviewer as meeting the DSM IV threshold (i.e., linial severity rating & 4). Agreement on the linial severity ratings of additional and lifetime diagnoses was not required, and the trainee was not required to math with the interviewer on diagnoses not formally assessed by the ADIS-IV-L (e.g., sexual disorders, eating disorders). Interviews were lassified as failing toward ertifiation when the trainee was rated as having ommitted one or more administration errors, regardless of whether his or her diagnoses mathed those of the senior interviewer. Results Reliability of DSM-IV Diagnosti Categories Current diagnoses. Interrater reliability of DSM-IV diagnoses was alulated by kappa oeffiients using the formula presented in Fleiss, Nee, and Landis (979). Following the guidelines used in studies of the reliability of DSM-III-R anxiety and mood disorders (e.g., Di Nardo et al., 993; Mannuzza et al., 989), the standards used to interpret kappa oeffiients were as follows: exellent agreement (K ^.75), good agreement (.60 < K =.74), fair agreement (.40 s «.59), and poor agreement (K <.40). In Table we present reliability findings for urrent DSM-IV diagnoses. For purposes of omparison, the findings from our reliability study of DSM-III-R anxiety and mood disorders (Di Nardo et al., 993) are also provided in Table. Using the aforementioned standards, we found that all prinipal diagnoses evidened good or exellent reliability with the exeption of dysthymia (DYS), although the kappas for pani disorder (PD) and DYS should be interpreted autiously beause these ategories were assigned infrequently as prinipal diagnoses in the sample (ns = 4 and 5, respetively).

4 52 BROWN, Di NARDO, LEHMAN, AND CAMPBELL Table Diagnosti Reliability of Current DSM-IV Diagnoses (N = 362) and Current DSM-HI-R Diagnoses (N = 267) Prinipal diagnosis Prinipal or additional diagnosis uiagnosti ategory PD PDA PD & PDA Speifi phobia Soial phobia GAD OCD PTSD MOD DYS MOD & DYS K DSM-IV DSM-III-R" DSM-IV DSM-IH-R* n K n K n K n Note, n = number of ases in whih diagnosis was assigned by either or both raters; dashes indiate an insuffiient n to alulate kappa; PD = pani disorder; PDA = pani disorder with agoraphobia; GAD = generalized anxiety disorder, OCD = obsessive-ompulsive disorder; PTSD = posttraumati stress disorder; MDD = major depressive disorder; DYS = dysthymia. a Data are from Di Nardo, Moras, Barlow, Rapee, and Brown (993). With the exeption of soial phobia (SOC), whih ontinued to be assoiated with exellent interrater agreement, higher kappas were observed for all prinipal DSM-IV anxiety and mood disorders relative to reliability findings for the orresponding DSM- III-R ategories. The most substantial improvement (i.e., from fair to good reliability) was evident for the prinipal diagnoses of PD (from.43 in DSM-III-R to.72 in DSM-IV), generalized anxiety disorder (GAD; from.57 in DSM-III-R to.67 in DSM-IV); and mood disorders (ollapsing major depressive disorder [MDD] and DYS; from.46 in DSM-III-R to.72 in DSM-IV). However, z tests of the differential magnitude of these kappas did not reah statistial signifiane. We noted a similar pattern of results when examining any urrent linial disorder, ollapsing aross prinipal and additional diagnoses (see Table ). Exellent reliability was obtained for pani disorder with agoraphobia (PDA), obsessive-ompulsive disorder (OCD), SOC, and pani disorder ollapsing aross the presene or absene of agoraphobia (PD and PDA). The ategories assoiated with good reliability were speifi phobia (SPEC), GAD, and any mood disorder (MDD and DYS). Fair reliability was found for PD, MDD, and posttraumati stress disorder (PTSD); DYS ontinued to be assoiated with poor reliability. As was the ase for prinipal diagnoses only, we obtained higher kappas (albeit not statistially signifiant as evaluated by z tests) for all DSM-IV ategories relative to DSM-HI-R, with the exeption of DYS whih went from.35 to.3 and OCD, whih did not hange (K =.75 in both studies). Disorder types and speifiers. Most DSM-IV ategories inlude additional sublassifiations to indiate the nature, ourse, or severity of the disorder. The reliability of these subtypes and speifiers was examined for any urrent linial disorder (i.e., prinipal or additional diagnosis). We evaluated the interrater agreement of the speifi phobia types and the generalized type of soial phobia using the entire sample. For MDD and DYS, reliability of speifiers was examined in ases in whih both interviewers assigned the disorder at a linial or sublinial level (i.e., speifiers are only reorded when MDD or DYS is diagnosed). The results of these analyses are presented in Table 2. At the level of prinipal diagnosis, exellent reliability was obtained for Table 2 Diagnosti Reliability of Current DSM-IV Diagnosti Types and Speifiers Type or speifier Speifi phobia Animal Natural environment Blood, injury, or injetion Situational Other Soial phobia Generalized Major depressive disorder Single or reurrent Mild, moderate, or severe Chroni or nonhroni Dysthymia Early or late onset Prinipal diagnosis Prinipal or additional diagnosis K Note. Dashes indiate an insuffiient n to alulate kappa. " For analyses of speifi phobia and soial phobia types, n refers to the number of ases in whih the type was assigned by either or both raters in the total study sample (N = 362); for major depressive disorder and dysthymia, n refers to size of the subsample (i.e., number of patients assigned the disorder by both raters at the linial or sublinial level) used in the analysis of speifiers..55 «a

5 RELIABILITY OF DSM-IV ANXIETY AND MOOD DISORDERS 53 eah of the speifi phobia types, although these findings should be interpreted with aution given the small sample sizes assoiated with some analyses. For all but the other type, these estimates dereased when reliability was examined using any urrent linial disorder. Consistent with a previous finding using DSM-I-R definitions (K =.69; Mannuzza et al., 995), the generalized type of DSM-IV soial phobia evidened good reliability both as a prinipal diagnosis and as a urrent diagnosis at any linial level (KS =.73). Interrater agreement for the ourse and severity speifiers of MDD and DYS is also presented in Table 2. Whereas the ourse and onset speifiers for MDD and DYS were assoiated with fair to good reliability (range of KS =.46 to.67), poor reliability was found for the MDD severity speifier (KS =.30 and.36). Reliability for the early/late onset speifier of prinipal DYS and speifiers for other disorders (e.g., poor insight in OCD) ould not be estimated beause of the exessively low rate that either the diagnosis or speifier was assigned in the sample. Soures of unreliability. Fators ontributing to diagnosti disagreements were evaluated in urrent linial diagnoses (ollapsing prinipal and additional status). As an be seen in Table 3, the prevailing soures of unreliability differed substantially aross the anxiety and mood disorders. For instane, the majority of disagreements involving SOC, SPEC, and OCD (62% to 67%) entailed ases in whih one interviewer assigned the diagnosis at a linial level and the other rated the diagnosis as sublinial; for other ategories (e.g., PDA, GAD, MDD, DYS), this was a relatively rare soure of unreliability. Indeed, the "threshold" issue was the most ommon soure of disagreements for the diagnoses of SPEC and SOC. Differene in patient report was otherwise the most prevalent soure of unreliability, ranging from 22% in SPEC to 00% in PTSD. Differential aggregation of unreliability soures was found for hange in linial status as well; although a rare soure for other disorders, it aounted for 9 of the 53 (7%) MDD disagreements, onsistent with the episodi nature of this ondition. Considerable variability was also evident aross ategories for the frequeny with whih other disorders were involved in diagnosti disagreements. Whereas disagreements with other disorders were relatively unommon for SOC, OCD, and PTSD (8% to 3%), another linial diagnosis was involved in over half of the disagreements with DYS, PDA, MDD, and GAD (54% to 74%). Table 3 provides the speifi disorders that were involved in these disagreements for eah diagnosis. As an be seen in this table, disagreements entailing another linial diagnosis quite often involved disorders that had overlapping definitional features and that differed mainly in the duration or severity of symptoms (e.g., PD vs. PDA; SPEC vs. agoraphobia without a history of PD; MDD vs. DYS). In addition, this overlap was evident in disagreements involving anxiety disorder not otherwise speified (NOS) and depressive disorder NOS diagnoses. For example, a ategory frequently involved in disagreements with GAD was anxiety disorder NOS (GAD; n = 0), in whih one interviewer noted linially signifiant features of GAD (i.e., linial severity rating ^ 4) but judged that not all riteria for a formal DSM-IV GAD diagnosis had been met (e.g., number or duration of worries or assoiated symptoms). This was also the ase for the NOS diagnoses assoiated with disagreements in other disorders (e.g., in the two OCD disagreements involving another disorder, both were with anxiety disorder NOS [OCD]). 2 Consistent with prior evidene that mood disorders may pose the greatest boundary problem for GAD, 22 of the 35 GAD disagreements (63%) involving another diagnosis were with mood disorders (DYS = 0, MDD = 9, depressive disorder NOS = 2, bipolar = ). Conversely, although most MDD disagreements involved other diagnoses (34 of 53), rarely (n = 3) were these disagreements with anxiety disorders. Indeed, most MDD disagreements involved other mood disorders (depressive disorder NOS = 5, DYS = 2). As shown in Table 3, other mood disorders were the most frequent diagnoses involved in DYS disagreements as well, although disagreements with GAD were more ommon (n = 6). Lifetime diagnoses. In Table 4 we present findings for the reliability of lifetime diagnoses (i.e., ollapsing aross urrent and past diagnoses). Beause alohol and substane use disorders were assigned frequently as past diagnoses, it was possible to evaluate the reliability of these ategories (the reliability of urrent alohol and substane use disorders ould not be examined beause of a study exlusion riterion). Exellent reliability was obtained for PDA, pani disorder ollapsing aross the presene or absene of agoraphobia (PD and PDA), OCD, alohol abuse or dependene, and substane abuse or dependene. SPEC, SOC, GAD, PTSD, MDD, and any mood disorder (MDD and DYS) were assoiated with good reliability. Fair reliability was found for PD. The lifetime diagnosis of DYS evidened poor interrater agreement. Reliability and Struture of DSM-IV Dimensional Features Data redution and fator analysis. We examined the interrater reliability of the dimensional ratings of DSM-IV anxiety and mood disorders features using the Boston University sample (n = 292). Prior to onduting reliability analyses, the ratings from eah ADIS-IV-L setion were submitted to fator analysis to provide an empirial basis for the formation of omposite sores (prinipal-omponents extration with oblique rotation, when needed). 3 In most instanes, unidimensional solutions were obtained, and these ADIS-IV-L setions were sored aordingly. However, analyses of OCD and SPEC ratings produed multifatorial strutures. Consistent with prior evidene of the multidimensionality of these symptoms (e.g., Summerfeldt, Rihter, Antony, & Swinson, 999), a three-fator solution was obtained for persistene and distress ratings of the nine types of OCD obsessions (three items eah): (a) ontamination, doubting, aidental harm to others; (b) aggressive and nonsensial impulses and sexual 2 To foster the desriptiveness of the anxiety disorder NOS and depressive disorder NOS ategories, a diagnosti on vention in our enter is to speify (in parentheses) the formal DSM-IV ategory to whih the NOS diagnosis is losest; for example, depressive disorder NOS (DYS) would be assigned in a ase in whih linially signifiant features of DYS are present (i.e., linial severity rating S4) but one or more of the DSM-IV riteria for DYS are not met (e.g., duration of slightly less than 2 years). 3 Analyses were limited to the Boston University subsample to ensure omplete data for all ontributing ases (i.e., ratings from the MDD, DYS, and assoiated symptoms of GAD setions were olleted on a listwise basis in the Boston University sample only; see Method setion). For the sake of brevity, details on the ondut and results of these fator analyses have been omitted from this report. A full desription of fator analyti results and a omprehensive list of ADIS-IV-L ratings are available by written request to Timothy A. Brown.

6 54 BROWN, Di NARDO, LEHMAN, AND CAMPBELL S. a 5* S ^ ^ 3 3 "S ^U f 3 a k. a, k. a 3 S ^Cj r*! o C Table 3 Fators Contributing to Dia^.«o % 5 03 S S U O O 03 < CL. a- j- o B Cu 2 2 u S IX Disorder or soui9 I ^O & Vi ON CSt-VOVO -H ( N O! TJ- in "v <N ^ ^ " "! in o in *n \ ^ "~~! ^-? O ^ ^ ^ "^ """! wa N 2 -< rt <* gl^ <* ~8 O CS 0 s ) O O ^ *O *"" "-H in i i i O ' ' " ' CU ~ OJ OO OO I I I I 2 I =a I rn r-j m ^o o m s i-t. <-s l l j O I O CS ON OO ^ O j I l l - t ^ *2 <>i ^ ^ ~, i i ON o s o ' ino TJ- ' o ON r-- -^ p- o JO' es o <N Oqqr^' o >n * o "" ^ CM ^ ^ ^ m b» s --H m o 3 2 P-r^S ^ oi2osos ffi T f i n I - H o)-h ^ m m mv~i ~- ^ 2 S S g2 3 TJ- inoo^n m -H G O I GO O O ' m m ( I O mtt^mrs.2 2 o T3.«" S «"o 'S 2 ~ 0 8 g g>? S J & ^ i ^» " i - ^, i g 3 - a 0 g f S t 3 0 o? f2 II II -g "o - g f Sr s 0 < «.S^.o ^ ii T3 ^ ^ S '! J i.^ g.g & 2 g"^.a 'S S, g ^ -3.5 S o> ii %r ^ 3 «ofg t l H o < - "o <f.2 B < - _ Z,3 ra u3 i- u i-. - o ^ S^ y «^J C " T) 0.E 3 p Ml HI! >»- i- -n «C 9 ' a o t g ^ II rt C '^ rt ' II ft. u ^ g H3 B. O DH W C?? OT a z.is S g g.2 «> to ^ -^ o o.3 'C t i 2 " II O *-i aj g j- WO ot*-g o S <a" "*> ^ 2 O' 0 'S. a " S, «s e ii g E ^ «o _ -r; S >i'«.2 O"5O > ) I/.2 t/ S >^ ^ > - ' -b S is u^ffi ll., ^ o «- M'C f & -a 5 03 S 'S.S g u S^-o ^ S a & S Slg^f ill fi el! ^ll?l on^3^ gu"., <u 'o.spgj&o^^ ^ «5 l-i Cfl - *K "O C! 5 <~i. to i~i ^4 tld tj) ^ ^ U 3 O g _C fa rt 'J5» 3 *J jg S OH -^ "o fis^s^^-ow ^!!! s s ^ «f i".s s ^ b" s s g-i'.s'o ^, 0~2 )<D 'i-g <s"3 M ^ ^ 3 JJ U -g o 'B S *0 aj ^ pj g fs^i s s««s b-l^ill.ii - h""-d 5 Dnft "H2'o ;^--ss O S C ^ ria^ Oi S ' a S,'^ " ^ "i 'u "S 'u o ^ j: ^ 3 oo D -g 't3 <u i (u ~' 3 3s2tg^.^ "S" o J::: '> ^ lleu«'"3o ob'o J= yvo/e. pro. = proportion; PD obsessive-ompulsive disorder; NOS = anxiety disorder not oth somatization disorder. Dashes ir " Cases in whih both raters re "Threshold" was the primary sou the other unreliability soures (e. disagreement pertained to a hour (hroni) versus DYS; that is, v,

7 RELIABILITY OF DSM-IV ANXIETY AND MOOD DISORDERS 55 Table 4 Diagnosti Reliability of Lifetime DSM-IV Diagnoses (N = 362) Lifetime diagnosis K n Pani disorder (PD) Pani disorder with agoraphobia (PDA) PD & PDA Speifi phobia Soial phobia Generalized anxiety disorder Obsessive-ompulsive disorder Posttraumati stress disorder Major depressive disorder (MOD) Dysthymia (DYS) MOD & DYS Alohol abuse or dependene Substane abuse or dependene Note, n = number of ases in whih diagnosis was assigned by either or both raters. thoughts or impulses; and () nonsensial thoughts/images, horrifi images, and religious/satani thoughts/impulses. A two-fator solution was obtained for the frequeny ratings of six OCD ompulsions. This struture entailed (a) the five ompulsions of heking, washing, adhering to rules or sequenes, internal repetition, and ounting; and (b) the single ompulsion of hoarding (f. Baer, 994). A four-fator solution was obtained for the fear ratings of 7 SPEC objets and situations: (a) blood/injury/injetion (6 items: blood from ut, reeiving injetions, having blood drawn either in self or others); (b) situational (5 items: elevators/enlosed plaes, air travel, driving, storms, heights); () Illness (3 items: vomiting, ontrating an illness, hoking); and (d) animals or water (2 items). However, beause animal fears and water fears were quite modestly orrelated (r =.5) and beause there was not a lear oneptual basis for ollapsing these ratings, they were evaluated separately in reliability analyses. In addition, fear of dental or medial proedures did not have a salient loading on any fator and was thus analyzed separately. Interrater reliability of ADIS-FV-L dimensional ratings. In Table 5 we provide reliability estimates (Pearson rs) for dimensional ratings of DSM-IV anxiety and mood disorder features. Although all are inluded for informational purposes, the omposite sores that pertained to different parameters of the same items were highly overlapping. Speifially, the following interorrelations were noted: (a) soial phobia fear versus avoidane ratings (r =.95), (b) speifi phobia fear versus avoidane ratings (range of rs =.76 to.90), () exessiveness versus unontrollability of GAD worry (r =.9), and (d) persistene or distress versus resistane of OCD obsessions (range of rs =.85 to.94). Aeptable interrater reliability was found for the majority of the various dimensional ratings. In most ases, the lowest estimates were for single-item ratings suh as speifi phobia avoidane of dental or medial proedures (.4) and avoidane of water (.48). The findings from reliability analyses of the 9-point (0-8) ADIS-IV-L linial severity rating for eah disorder are also shown in Table 5. uite favorable reliability was obtained for the linial severity ratings of most disorders. However, onsistent with findings at the diagnosti level, reliability of the DYS linial severity rating was low (r =.36). Disussion Diagnosti Reliability of Current and Lifetime DSM-IV Anxiety and Mood Disorders Colletively, these findings suggest that most urrent disorders are assoiated with good to exellent interrater agreement. 4 For example, all prinipal diagnosti ategories exept DYS evidened good to exellent reliability. In omparison with our DSM-III-R reliability study (Di Nardo et al., 993), improved reliability was noted for the vast majority of DSM-IV disorders, and no DSM-IV ategory was assoiated with a markedly lower reliability estimate. Diagnoses showing the most improved reliability were PD and GAD. As was the ase for urrent diagnoses, good to exellent reliability was found for the majority of lifetime anxiety and mood disorders. Interestingly, exellent interrater agreement was obtained for the alohol and substane use disorders (KS =.83 and.82, respetively), indiating the potential utility of the ADIS- IV-L to provide reliable DSM-IV diagnosis of these onditions. The improved reliability of GAD is partiularly enouraging beause this ategory was in jeopardy of being removed from DSM-IV, in part beause of the poor to fair reliability of its DSM-III-R definition. This improvement ould perhaps be attributed to the revised definition of GAD in DSM-IV, whih emphasizes the unontrollable nature of worry and the assoiated symptoms of tension and negative affet. However, GAD diagnosti disagreements frequently involved the mood disorders (47%). 5 This is onsistent with prior evidene (e.g., Brown et al., 998; Starevi, 995) that the mood disorders pose a more signifiant boundary issue for GAD than do other anxiety disorders. In future researh, it would be important to examine the disriminant validity of GAD and mood disorders and determine if the diagnosti definition of GAD ould be further refined to foster its distintion from these onditions. Also noteworthy is the finding that differene in patient report was rated the most ommon soure of GAD disagreement (55%). This finding ould also be refletive of limitations in the diagnosti riteria. Reliable diagnosis of GAD requires onsistent self-report of many subjetive features (e.g., number and severity of worry areas and physial symptoms) and their onset and duration in relation to other onditions (e.g., mood disorders). Inonsisteny in suh reports ould be indiative of vagueness of these diagnosti features and patients' diffiulty differentiating them from other disorders. Bearing on this point, previous researh has shown that disorders assoiated with lear behavioral markers (e.g., OCD with ompulsions, and situational avoidane in PDA, SOC, or SPEC) are assoiated with higher reliability than disorders without suh features (e.g., PD, GAD, 4 One ould argue that the present rates of interrater agreement represent the upper limit of potential reliability estimates for these disorders given aspets of the study methodology suh as use of highly trained interviewers and the speialized anxiety and mood disorders setting (i.e., diagnosti reliability might be lower in primary linial settings that often entail patient populations of a wider range of disorders, less strutured linial assessments, et.). 5 It is noteworthy than none of the GAD disagreements involved OCD (or vie versa) despite previous onerns about boundary problems with exessive worry and obsessions (Brown, Moras, Zinbarg, & Barlow, 993; Turner, Beidel, & Stanley, 992).

8 56 BROWN, Di NARDO, LEHMAN, AND CAMPBELL Table 5 Interrater Reliability ofadis-iv-l Dimensional Ratings of DSM-IV Disorder Features Feature or rating Pani disorder/agoraphobia Number of pani attaks (past month).58 Fear of pani attaks (past month).53 Agoraphobi avoidane.86 Clinial severity rating.83 Soial phobia Situational fear.86 Situational avoidane.86 Clinial severity rating.80 Generalized anxiety disorder Exessive worry.73 Unontrollability of worry.78 Assoiated symptoms.83 Clinial severity rating.72 Obsessive-ompulsive disorder Obsessions: persistene distress Doubting, ontamination, aidental harm.75 Impulses (aggressive, sexual, nonsensial).68 Other (religious, horrifi, nonsensial thoughts).78 Obsessions: resistane Doubting, ontamination, aidental harm.76 Impulses (aggressive, sexual, nonsensial).43 Other (religious, horrifi, nonsensial thoughts).72 Compulsions Compulsion frequeny.79 Hoarding frequeny.58 Clinial severity rating.84 Speifi phobia Situational fear Blood, injury, injetion.77 Situational.73 Vomiting, hoking, ontrating an illness.63 Animals.64 Water.54 Dental or medial proedures.53 Situational avoidane Blood, injury, injetion.73 Situational.73 Vomiting, hoking, ontrating an illness.66 Animals.72 Water.48 Dental or medial proedures.4 Clinial severity rating.75 Major depression Key symptoms.74 Clinial severity rating.65 Dysthymia Key symptoms.78 Clinial severity rating.36 Any mood disorder (major depression or dysthymia) Clinial severity rating.69 Note. ADIS-IV-L = Anxiety Disorders Interview Shedule for DSM- IV: Lifetime version. N = 292 for all analyses exept for analyses of pani disorder/agoraphobia number of pani attaks, fear of pani attaks, and agoraphobi avoidane ratings (ns = 97). For all rs, p <.00. and OCD without ompulsions; Chorpita, Brown, & Barlow, 998). As in previous studies (Di Nardo et al., 993; Williams et al., 992), the urrent and lifetime diagnosis of DYS possessed poor reliability, further alling into question the utility of this ategory r as urrently defined. Although the potential overlap of DYS and GAD is apparent (i.e., both disorders onstitute hroni symptoms of negative affet), it is noteworthy that the vast majority of DYS disagreements involved other mood disorders. This was also true for MDD disagreements in whih the anxiety disorders were rarely involved. This suggests that boundary issues within the mood disorders are a primary soure of unreliability, often pertaining to limitations of the ategorial approah suh as differentiating (a) DYS from hroni MDD and (b) MDD and DYS from depressive disorder NOS. This also aounts for the findings of higher reliability when MDD and DYS were ollapsed into one ategory than when they were analyzed as separate ategories (see Tables and 4). Unreliability Due to Diagnosti Threshold Issues Although a similar pattern of reliability estimates was obtained when any urrent diagnoses were examined (i.e., ollapsing prinipal and additional diagnoses), interrater agreement of PD, OCD, and SPEC evidened a marked deline relative to their estimates as prinipal diagnoses. Inspetion of the soures of unreliability indiated that these ategories were the most prone to disagreement involving diagnosti thresholds that is, both interviewers reorded key features of the disorders but disagreed on the presene of suffiient impairment and distress to assign a formal DSM-IV diagnosis (e.g., this issue was responsible for 62% of SPEC disagreements). This was a strong ontributing fator to redued reliability of PD, OCD, and SPEC beause additional diagnoses were more suseptible to the threshold issue than were prinipal diagnoses (i.e., by definition, a prinipal diagnosis is the disorder assoiated with the highest degree of distress or interferene). Similarly, although exellent reliability was evident for the five SPEC types as prinipal diagnoses, these estimates delined for most SPEC types when ollapsing prinipal and additional diagnoses. This again was attributable mainly to higher rates of diagnosti threshold disagreements, although ertain SPEC types were more affeted by this issue (i.e., animal, natural environment, blood/injury/injetion); thus, defining the boundary of linially signifiant interferene and distress may be more diffiult for some forms of SPEC (e.g., although marked impairment or distress may be learly indiated in Situational fears suh as driving, it may be less apparent in fears of things suh as animals, heights, et., whih the person rarely enounters or an avoid without onsiderable lifestyle impat). The diagnosti threshold issue also illustrates the problem of measurement error introdued by imposing ategorial utoffs (i.e., DSM-IV riteria for the presene or absene of a disorder) on diagnosti features that operate largely in a ontinuous fashion (e.g., number, severity, and duration of symptoms and degree of distress). Evaluation of soures of unreliability suggests several other instanes in whih this ourred. Many of the diagnosti disagreements assoiated with GAD, MDD, and DYS involved anxiety disorder NOS and depressive disorder NOS. This indiates that both interviewers agreed on the presene of linially signifiant features of the disorder in question (linial severity ratings S: 4), but that one interviewer did not assign a formal anxiety or mood disorder diagnosis beause of subthreshold patient report of the number or duration of symptoms (beause of inonsistent report, hange in linial status, et.). Another example of this

9 RELIABILITY OF DSM-IV ANXIETY AND MOOD DISORDERS 57 problem pertained to the severity speifiers for MDD. Whereas dimensional ratings of the severity of MDD features were quite reliable (r =.74; Table 5), the DSM-IV ategorial speifiers of MDD severity evidened poor reliability (KS =.30 and.36; Table 2). Beause of the measurement error, loss of information, and validity problems assoiated with the purely ategorial approah to diagnosti lassifiation in DSM-IV, researhers have alled for inorporation of dimensional omponents in future nosologial systems (e.g., Blashfield, 990; Brown, in press; Franes, Widiger, & Fyer, 990). Indeed, favorable reliability was found for most omposite dimensional ratings of disorder features and for single ratings suh as the linial severity ratings (Table 5). These findings are noteworthy in view of the wide use of these measures as indexes of treatment outome (e.g., Borkove & Costello, 993; Brown & Barlow, 995) and as indiators in studies of the nature of emotional disorders (e.g., Brown et al, 998). Although intended to provide psyhometri justifiation for omposite soring, the results of the fator analyses of the ADIS-IV-L dimensional ratings may have impliations for the typology of some disorders. For example, analysis of fear ratings of 7 speifi phobia situations did not support the presene of a distint fator representing natural environment-type phobias. Instead, suh fears either tended to be assoiated with situational fears (heights, storms) or failed to aggregate saliently with any other fear (water). This result ould be interpreted to support prior arguments and preliminary findings that some natural-environment-type fears (e.g., heights) are better onstrued as situational-type phobias (Antony et al., 997). Summary and Conlusions The urrent findings provide support for the reliability of most DSM-IV emotional disorders as assessed by the ADIS-IV-L and eluidate soures of error in the diagnosis of these onditions. However, these findings learly show that the DSM-IV anxiety and mood disorders were differentially affeted by the various soures of unreliability. Besides MDD and DYS (whose disagreements frequently involved eah other), only GAD and SPEC had onsiderable rates of disagreements involving other diagnosti ategories (mood disorders in GAD, agoraphobia in SPEC), whih might suggest that these disorders are more prone to error assoiated with overlapping key or assoiated features. For many ategories (e.g., SOC and OCD), disagreements rarely involved other disorders and were primarily due to problems in defining and applying a ategorial threshold to the lassifiation of the number, severity, or duration of symptoms (e.g., disagreements on linial vs. sublinial diagnoses and disagreements involving NOS diagnoses). Although the linial versus sublinial issue was less relevant in reliable diagnosis of PDA, GAD, MDD, and DYS, unreliability related to ategorial threshold was evident in these disorders by the high inidene of disagreements with NOS diagnoses and MDD versus DYS. Thus, the high rate of disagreements involving thresholds and NOS diagnoses indiated that in many ases interviewers onurred on the presene of the features of a given disorder; however, unreliability was introdued through the diffiulties in applying the DSM-IV ategorial utoff to these features. These data support the need for ontinued researh that may ultimately and unequivoally doument the importane of dimensionally based assessment systems in improving our formal approahes to the lassifiation of psyhologial disorders. Referenes Abel, J. L., & Borkove, T. D. (995). Generalizability of DSM-III-R generalized anxiety disorder to proposed DSM IV riteria and rossvalidation of proposed hanges. Journal of Anxiety Disorders, 9, Amerian Psyhiatri Assoiation. (987). Diagnosti and statistial manual of mental disorders (3rd ed., rev,). Washington, DC: Author. Amerian Psyhiatri Assoiation. (994). Diagnosti and statistial manual of mental disorders (4th ed.). Washington, DC: Author. Antony, M. M., Brown, T. A., & Barlow, D. H. (997). Heterogeneity among speifi phobia types in DSM-IV. Behaviour Researh and Therapy, 35, Antony, M. M., Moras, K., Meadows, E. A., Di Nardo, P. A., Uteh, J. E., & Barlow, D. H. (994). The diagnosti signifiane of the funtional impairment and subjetive distress riterion: An illustration with the DSM-IH-R anxiety disorders. Journal of Psyhopathology and Behavioral Assessment, 6, Baer, L. (994). Fator analysis of symptoms subtypes of obsessiveompulsive disorder and their relation to personality and ti disorders. Journal of Clinial Psyhiatry, 55, Barlow, D. H., Brown, T. A., & Craske, M. G. (994). Definitions of pani attaks and pani disorder in DSM-IV: Impliations for researh. Journal of Abnormal Psyhology, 03, Blashfield, R. K. (990). Comorbidity and lassifiation. In J. D. Maser & C. R. Cloninger (Eds.), Comorbidity of mood and anxiety disorders (pp. 6-82). Washington, DC: Amerian Psyhiatri Press. Borkove, T. D. (994). The nature, funtions, and origins of worry. In G. Davey & F. Tallis (Eds.), Worrying: Perspetives on theory, assessment, and treatment (pp. 5-33). New York: Wiley. Borkove, T. D., & Costello, E. (993). Effiay of applied relaxation and ognitive behavioral therapy in the treatment of generalized anxiety disorder. Journal of Consulting and Clinial Psyhology, 6, Brawman-Mintzer, O., Lydiard, R. B., Emmanuel, N., Payeur, R., Johnson, M., Roberts, J., Jarrell, M. P., & Ballanger, J. C. (993). Psyhiatri Comorbidity in patients with generalized anxiety disorder. Amerian Journal of Psyhiatry, 50, Brown, T. A. (996). Validity of the DSM-UI-R and DSM-IV lassifiation systems for anxiety disorders. In R. M. Rapee (Ed.), Current ontroversies in the anxiety disorders (pp. 2-45). New York: Guilford Press. Brown, T. A. (in press). The lassifiation of anxiety disorders: Current status and future diretions. In D. J. Stein & E. Hollander (Eds.), Textbook of anxiety disorders. Washington, DC: Amerian Psyhiatri Press. Brown, T. A., & Barlow, D. H. (992). Comorbidity among anxiety disorders: Impliations for treatment and DSM-IV. Journal of Consulting and Clinial Psyhology, 60, Brown, T. A., & Barlow, D. H. (995). Long-term outome in ognitivebehavioral treatment of pani disorder: Clinial preditors and alternative strategies for assessment. Journal of Consulting and Clinial Psyhology, 63, Brown, T. A., Barlow, D. H., & Liebowitz, M. R. (994). The empirial basis of generalized anxiety disorder. Amerian Journal of Psyhiatry, 5, Brown, T. A., Chorpita, B. F., & Barlow, D. H. (998). Strutural relationships among dimensions of the DSM-IV anxiety and mood disorders and dimensions of negative affet, positive affet, and autonomi arousal. Journal of Abnormal Psyhology, 07, Brown, T. A., Marten, P. A., & Barlow, D. H. (995). Disriminant validity of the symptoms onstituting the DSM-III-R and DSM-IV assoiated

10 58 BROWN, Dl NARDO, LEHMAN, AND CAMPBELL symptom riterion of generalized anxiety disorder. Journal of Anxiety Disorders, 9, Brown, T. A., Moras, K., Zinbarg, R. E., & Barlow, D. H. (993). Diagnosti and symptom distinguishability of generalized anxiety disorder and obsessive-ompulsive disorder. Behavior Therapy, 24, Chorpita, B. F., Brown, T. A., & Barlow, D. H. (998). Diagnosti reliability of the DSM-III-R anxiety disorders: Mediating effets of patient and diagnostiian harateristis. Behavior Modifiation, 22, Clark, L. A., & Watson, D. (99). Tripartite model of anxiety and depression: Psyhometri evidene and taxonomi impliations. Journal of Abnormal Psyhology, 00, Costello, C. G. (992). Researh on symptoms versus researh on syndromes: Arguments in favour of alloating more researh time to the study of symptoms. British Journal of Psyhiatry, 60, Di Nardo, P. A., & Barlow, D. H. (988). Anxiety Disorders Interview Shedule Revised (ADIS-R). Albany, NY: Graywind. Di Nardo, P. A., Brown, T. A., & Barlow, D. H. (994). Anxiety Disorders Interview Shedule for DSM-IV: Lifetime version (ADIS-V-L). San Antonio, TX: Psyhologial Corporation. Di Nardo, P. A., Moras, K., Barlow, D. H., Rapee, R. M., & Brown, T. A. (993). Reliability of DSM-III-R anxiety disorder ategories using the Anxiety Disorders Interview Shedule Revised (ADIS-R). Arhives of General Psyhiatry, 50, Fleiss, J. L., Nee, J. C. M., & Landis, J. R. (979). Large sample variane of kappa in the ase of different sets of raters. Psyhologial Bulletin, 86, Franes, A., Widiger, T., & Fyer, M. R. (990). The influene of lassifiation methods on omorbidity. In J. D. Maser & C. R. Cloninger (Eds.), Comorbidity of mood and anxiety disorders (pp. 4-59). Washington, DC: Amerian Psyhiatri Press. Mannuzza, S., Fyer, A. J., Martin, L. Y., Gallops, M. S., Endiott, J., Gorman, I. M., Liebowitz, M. R., & Klein, D. F. (989). Reliability of anxiety assessment: I. Diagnosti agreement. Arhives of General Psyhiatry, 46, Mannuzza, S., Shneier, F. R., Chapman, T. F., Liebowitz, M. R., Klein, D. F., & Fyer, A. J. (995). Generalized soial phobia: Reliability and validity. Arhives of General Psyhiatry, 52, Marten, P. A., Brown, T. A., Barlow, D. H., Borkove, T. D., Shear, M. K., & Lydiard, R. B. (993). Evaluation of the ratings omprising the assoiated symptom riterion of DSM-III-R generalized anxiety disorder. Journal of Nervous and Mental Disease, 8, Starevi, V. (995). Pathologial worry in major depression: A preliminary report. Behaviour Researh and Therapy, 33, Stein, M. B., Walker, J. R., & Forde, D. R. (994). Setting diagnosti thresholds for soial phobia: Considerations from a ommunity survey of soial anxiety. Amerian Journal of Psyhiatry, 5, Summerfeldt, L. I., Rihter, M. A., Antony, M. M., & Swinson, R. P. (999). Symptom struture in obsessive-ompulsive disorder: A onfirmatory fator-analyti study. Behaviour Researh and Therapy, 37, Turner, S. M., Beidel, D. C., & Stanley, M. A. (992). Are obsessional thoughts and worry different ognitive phenomena? Clinial Psyhology Review, 2, Williams, J. B. W., Gibbon, M., First, M. B., Spitzer, R. L., Davies, M., Borus, J., Howes, M. J., Kane, I., Pope, H. G., Rounsaville, B., & Witthen, H. (992). The Strutured Clinial Interview for DSM-III-R (SCID): II. Multisite test-retest reliability. Arhives of General Psyhiatry, 49, Reeived February 7, 2000 Revision reeived August, 2000 Aepted August 3, 2000